MixMatch-pytorch issues

At what point does the program stop running

1

I'm running this code and it's working great! But I noticed that epoch has 1024 and I've run 110 of them now, each taking about a minute, will the program...

Enderfga

Update to work with pytorch 1.11.0

1

Previously running this on pytorch 1.11.0 failed with `RuntimeError: "nll_loss_forward_reduce_cuda_kernel_2d_index" not implemented for 'Int'`. After converting training labels to long (which is not for some reason a requirement), I ran...

aleloi

I use mixmatch to train my own data set, which is simpler than cifar10 image. After running several epoch, the loss function of training and trainstats will become larger, and the accuracy of test set will be reduced

3

ghost

Allow training on customized dataset

I revised the original train.py. I am a new guy in this field, therefore any suggestions are welcomed! Check README_customized_data.md to see the training process. The updated train_SSL.py allows training...

duyao-art

why use .detach() here

Hi Thanks for the great repo. I'm wondering in the line below, why do we need .detach(), isn't that the targets_u is already in the scope of torch.no_grad()? https://github.com/YU1ut/MixMatch-pytorch/blob/cc7ef42cffe61288d06eec1428268b384674009a/train.py#L235

hzhz2020

Unlabelled batch size

5

According to the paper, there's a set of labelled and unlabelled datapoints that are mixed. In the code nevertheless there are two unlabelled sets, u1 and u2. Both of them...

AlexTS1980

interleave function

1

Hi, Thanks for the great repository! Please tell what is interleave function doing in the code? It is not clear from the code itself and there is nothing mentioned about...

07Agarg

Performance declining by replacing data augmentation with pyotrch function.

3

Hello, I replace your data augmentation function (`RandomPadandCrop`, `RandomFlip`) with pytorch augmenation function (`RandomCrop`, `RandomHorizontalFlip`). But the performance declines from 94% to 66%. My data augmantion code is as follow:...

yichuan9527